Each of the traditional modes of data collection, whether face-to-face, telephone, or mail—has a rhythm and character that sets it apart from the others. The selection of mode has a significant effect on how samples are drawn, how individual questions get worded, and how items are ordered to form questionnaires. These differences often influence how people provide answers to individual questions, and even whether they respond at all to a particular survey. It would be surprising if the introduction of surveys by the Internet, did not also bring with it many new, often unintended, data collection practices that set it apart from other modes of surveying and which also influence data quality. Our purpose in this presentation is to identify a number of significant changes in data collection practices that are consequences of surveys on the Internet that are based upon periodic surveys of random samples of students atWashington State University conducted by the authors. These surveys are designed to provide useful longitudinal information to University leadership for tracking how students are responding to the education being provided by our University. In this paper we discuss a wide variety of coverage, measurement and non-response issues that require us to approach the web surveying process differently than when we relied only on telephone and/or mail systems. The changes include coverage issues, e.g. not being able to get a single list of postal, telephone, or web addresses, because of changes in what address information students are required to provide to the university and the change from mandatory updating to making it voluntary as to which address is available for university use. Another change we discuss is the benefit of using contacts by an alternative mode (mail) to deliver meaningful incentives in order to achieve high web survey response rates. Based upon extensive experimentation which shows that differences in the visual layout of questionnaires influence how people respond to web survey questions we discuss the use of procedures such as Cascading style sheet construction and other measures in order to control the visual aspect of questions seen by respondents in an effort to provide the same stimulus to each respondent. Another critical change has to do with the development of monitoring systems to quickly analyze and respond to respondent problems and concerns that are critical in a web context for converting likely refusals to completes. These are examples of the topics to be discussed. The learnings from these periodic surveys have provided a basis for developing a web survey system now used by the Social and Economic Sciences Research Center that we now employ in a general way for doing web surveys of many other populations. In the proposed paper we will describe in detail both the problems and solutions associated with maintaining high quality survey data collection processes, making specific reference to how the web has required us to change long-standing data collection practices that previously supported our telephone and mail data collection systems.